The Sunny Side of Fairness - Preference for Fairness Activates Reward Circuitry (and Disregarding Unfairness Activates Self-Control Circuitry)

Page created by Gordon Warren
 
CONTINUE READING
The Sunny Side of Fairness - Preference for Fairness Activates Reward Circuitry (and Disregarding Unfairness Activates Self-Control Circuitry)
PS YC HOLOGICA L SC IENCE

Research Article

The Sunny Side of Fairness
Preference for Fairness Activates Reward Circuitry (and
Disregarding Unfairness Activates Self-Control Circuitry)
Golnaz Tabibnia, Ajay B. Satpute, and Matthew D. Lieberman

University of California, Los Angeles

ABSTRACT—Little    is known about the positive emotional                  desirable for the recipient than unfair outcomes in everyday life,
impact of fairness or the process of resolving conflict be-               it is difficult to distinguish the desire for fairness from the desire
tween fairness and financial interests. In past research,                 for material gain. Bilateral bargaining games, such as the ulti-
fairness has covaried with monetary payoff, such that the                 matum game, allow these two potential motives to be examined
mental processes underlying preference for fairness and                   separately. The results of studies using the ultimatum game indi-
those underlying preference for greater monetary outcome                  cate that people are sensitive to fairness over and above its conse-
could not be distinguished. We examined self-reported hap-                quences for material gain (Güth, Schmittberger, & Schwarze,
piness and neural responses to fair and unfair offers while               1982). Although there is evidence that receiving an unfair pro-
controlling for monetary payoff. Compared with unfair                     posal is associated with negative emotional responses (Sanfey,
offers of equal monetary value, fair offers led to higher                 Rilling, Aronson, Nystrom, & Cohen, 2003), no study on economic
happiness ratings and activation in several reward regions                decision making has examined whether a fair proposal produces
of the brain. Furthermore, the tendency to accept unfair                  positive emotional responses beyond those associated with the
proposals was associated with increased activity in right                 material gain itself.
ventrolateral prefrontal cortex, a region involved in emo-                    To examine the emotional response associated with fair
tion regulation, and with decreased activity in the anterior              treatment, we conducted two ultimatum-game experiments. In
insula, which has been implicated in negative affect. This                this game, one player proposes how to split a given sum of money,
work provides evidence that fairness is hedonically valued                the stake, and another player responds. If the responder accepts,
and that tolerating unfair treatment for material gain                    each player keeps the amount allocated by the proposer. If the
involves a pattern of activation resembling suppression of                responder rejects the offer, neither player receives any money.
negative affect.                                                          Numerous studies using the ultimatum game have shown that
                                                                          responders do not maximize material utility by accepting every
Anyone who has watched children negotiate how to share a piece            offer, but rather tend to reject offers below 20% of the stake
of cake knows that humans are exquisitely sensitive to fairness.          (Camerer & Thaler, 1995), even when there will be no future
Although economic models of decision making have tradition-               interactions with the partner (Güth et al., 1982).
ally assumed that individuals are motivated solely by material                In a neuroimaging study of the ultimatum game, Sanfey et al.
utility (e.g., financial payouts) and are not directly affected by        (2003) observed that being treated unfairly is associated with
social factors such as fairness (Camerer, Loewenstein, & Prelec,          a negative emotional response, inferred from anterior insula
2005; Kahneman, Knetsch, & Thaler, 1986), there is increasing             activation. They did not report what regions were more active
empirical evidence that fairness does play a role in economic             during fair than during unfair offers. Furthermore, because fair
decision making (Fehr & Schmidt, 1999; Sears & Funk, 1991).               offers (i.e., $5 out of $10) were always associated with higher
Fairness in economic-exchange tasks is typically defined as               monetary payoff than unfair offers (e.g., $2 out of $10), it is
the equitable distribution of an initial stake of money between           difficult to dissociate emotional response to fairness from emo-
two people. Because fair outcomes tend to be more materially              tional response to monetary payoff in their study. Hence, it is
                                                                          unclear from these data whether fair treatment is rewarding, in
                                                                          addition to unfair treatment being aversive.
Address correspondence to Golnaz Tabibnia, The Semel Institute for
                                                                              Research on social justice suggests that seeking justice is a
Neuroscience and Human Behavior, University of California, Los
Angeles, 760 Westwood Plaza, C8-532, Los Angeles, CA 90095-1759,          basic human impulse (i.e., the justice motive; Tyler, 1991), pos-
e-mail: golnaz@ucla.edu.                                                  sibly rooted in a basic social motivation to be accepted (Bau-

Volume 19—Number 4                            Copyright r 2008 Association for Psychological Science                                       339
The Sunny Side of Fairness - Preference for Fairness Activates Reward Circuitry (and Disregarding Unfairness Activates Self-Control Circuitry)
Fairness Is Rewarding

meister & Leary, 1995). Perceived fair treatment from public                       We also examined neural response during trials in which
institutions (e.g., court, police) has been associated with satis-              fairness and material outcome were at odds—that is, trials
faction beyond the effects of the material outcomes, such as                    on which the offers were unfair but financially desirable (e.g.,
sentencing (Tyler, 1984). Critically, studies examining the im-                 $8 out of $23). Thus, we examined the neural correlates of
pact of fairness on positive and negative emotions separately,                  the tendency to accept unfair offers. Two possibilities were in-
controlling for material outcomes, have found substantial in-                   vestigated. First, accepted unfair offers may activate reward
creases in self-rated positive emotions associated with fair treat-             circuitry to a greater extent than rejected unfair offers; such a
ment (De Cremer & Alberts, 2004; Hegtvedt & Killian, 1999).                     pattern would reflect enhanced desire to accept the offers. Sec-
   If being treated fairly is experienced as rewarding, then                    ond, emotion regulation may be engaged when unfair offers are
people should be happier with a fair offer than with an unfair                  accepted, which would diminish the anterior insula’s response
offer of the same monetary value. Similarly, brain regions as-                  and decrease the desire to reject the offer. In this case, one would
sociated with reward should be more active during fair than                     expect decreased activity in the anterior insula and increased
during unfair treatment, after controlling for material utility.                activity in a prefrontal region that has been associated with
These reward regions include the ventral striatum, the amygdala,                emotion regulation, such as the right ventrolateral prefrontal
ventromedial prefrontal cortex (VMPFC), orbitofrontal cor-                      cortex (right VLPFC; Hariri, Bookheimer, & Mazziotta, 2000;
tex (OFC), and midbrain dopamine regions (Cardinal, Parkin-                     Lieberman et al., 2007).
son, Hall, & Everitt, 2002; Trepel, Fox, & Poldrack, 2005).                        Participants in our studies played the role of responder. In
Although the amygdala has commonly been associated with fear                    Experiment 1, we measured emotional responses to each offer by
processes, activity in this structure, particularly on the left, has            obtaining self-ratings of happiness and contempt. In Experiment
also been associated with reward processes (Hommer et al.,                      2, we measured neural responses to fair and unfair offers using
2003; Zalla et al., 2000).                                                      functional magnetic resonance imaging (fMRI). Participants
   In order to control for material utility, we varied both the offer           were told that their decisions regarding four randomly selected
amount and the stake size across trials (see Fig. 1). On different              offers in the experiment would actually be implemented, such
trials, the same offer amount could represent a large percentage                that they and the proposers of those offers would be paid or not,
of the total stake (e.g., $7 out of $15), and therefore seem fair, or           according to the responses.
a small percentage of the total stake (e.g., $7 out of $23), and
therefore seem unfair. Differences in ratings of happiness or
reward activations observed in the comparison of such trials                                            EXPERIMENT 1
cannot be attributed to the magnitude of the monetary reward
and thus are reasonably attributed to fairness.                                 Method
                                                                                Participants and Task
                                                                                Twenty-nine undergraduates (average age 5 20.1 years; 18 fe-
                                                                                males, 11 males) participated after replying to a flyer indicating
                                                                                that they could earn up to $52 for participation. They were told
                                                                                that the proposers had submitted their offers already and would
                                                                                not be present. Actually, there were no real proposers. During
                                                                                the experiment, each offer was presented as follows (see Fig. 2):
                                                                                First, participants were shown the purported proposer. Then, the
                                                                                stake was indicated, followed by the offer. While the offer was
                                                                                displayed, participants could accept or reject it. After the ex-
                                                                                periment, all participants were debriefed, paid a total of $27,
                                                                                and entered in a lottery in which 4 participants were selected to
                                                                                receive an additional $25. Thus, all participants had a chance of
                                                                                winning ‘‘up to $52,’’ as advertised. Offers ranged from 5% to
                                                                                50% of the total stake size, and stakes ranged from $1 to $30. We
                                                                                selected particular offer values and then matched each with two
                                                                                stake sizes in order to obtain one low and one high ratio of offer to
Fig. 1. Illustration of the manipulation of material utility and fairness.
In the analysis of fairness preference, trials of equal material utility were
                                                                                stake size.
divided according to fairness (i.e., the ratio of the offer to the stake). In
this example, the offers in the top row are high-fairness offers, and those     Measures
in the bottom row are low-fairness offers, and each of two monetary             After playing the game, participants were asked to rate (1–7)
outcomes is presented in both a high-fairness offer and a low-fairness
offer. Across trials, high- and low-fairness offers had the same average        how much happiness and contempt they felt in response to each
material utility.                                                               of a preselected subset of 28 offers. This subset consisted of 14

340                                                                                                                                Volume 19—Number 4
Golnaz Tabibnia, Ajay B. Satpute, and Matthew D. Lieberman

                                                                                         $10
                                +                                                                                   $10

                                                                                                  $2

                                                                      $0     $5 $10 $15
                                                                                                 $0     $5 $10 $15 $20 $25 $30

                                Blank          Portrait              Stake                            Offer

                               0      0.5                     2.0              3.0                                            6.0
                                                                     Time (seconds)
           Fig. 2. Diagram illustrating the structure of each 6-s trial: The participant saw a fixation cross for 0.5 s, a picture of the
           purported proposer for 1.5 s, a display indicating the size of the stake for 1 s, and finally the offer for 3 s. The participant was
           given the final 3 s of each trial to respond, by pressing one button to ‘‘accept’’ and another to ‘‘reject’’ the offer. Then the
           next blank screen appeared for 0.5 s, and so forth.

fair offers ( 40% of the stake) and 14 unfair offers ( 20% of                ðHC̄ Þ was still associated with fairness (r 5 .38, prep > .95);
the stake) that were matched in material utility (e.g., $2 out of $4           however, contempt controlling for happiness (CH̄ ), was not (r 5
matched with $2 out of $10). In this subset of offers, the ratio of            .17, prep 5 .80).
offer to stake size ranged from 5 to 50% (average 5 28%); the
stakes ranged from $1 to $30 (average 5 $12.18).
                                                                               Fairness Predicts Happiness, Independently of Material Outcome
                                                                               Offer amount and emotion ratings were not strongly associated.
Results
                                                                               Happiness ratings of high-value offers (> $2) did not differ from
Fairness Predicts Happiness, Independently of Contempt                         those of low-value offers ( $2) of equal fairness, t(13) 5 1.16,
Happiness ratings were strongly associated with fairness (i.e.,                prep 5 .87, d 5 0.64. Although the correlation between offer
percentage of the stake size offered). Participants reported                   amount and happiness was marginally significant (r 5 .32,
greater happiness for fair offers ( 40%) than unfair offers                   prep > .95), offer amount did not predict HC (r 5 .28, prep > .92).
( 20%) of equal monetary value, t(13) 5 7.73, prep > .99,                     Similarly, contempt ratings of high-value offers did not differ
d 5 4.29. Similarly, there was a strong correlation between level              from those of low-value offers of equal fairness, t(13) 5 0.39,
of fairness and happiness ratings (r 5 .89, prep > .99). A com-                prep 5 .60, d 5 0.15. The correlations between offer amount and
plementary pattern was observed for contempt ratings. Partici-                 contempt (r 5 .20, prep 5 .85) and between offer amount and
pants reported greater contempt for unfair offers than for                     CH̄ (r 5 .11, prep 5 .71) were not significant.
fair offers of equal monetary value, t(13) 5 5.51, prep > .99,                    After controlling for the effect of offer amount, fairness still
d 5 3.06, and there was a strong correlation between level of                  predicted happiness (b 5 .87, prep > .99) and HC (b 5 .35,
fairness and contempt ratings (r 5 .82, prep > .99).                          prep > .95), and fairness still predicted contempt (b 5 .81,
   Given that happiness and contempt were correlated (r 5 .82,                prep > .99), but not CH (b 5 .18, prep 5 .82). Together, these
prep > .99), we examined the effect of fairness on happiness                   results indicate that fairness, independently of offer amount,
after partialing out contempt. Happiness controlling for contempt              predicts happiness, independently of contempt.

Volume 19—Number 4                                                                                                                                341
Fairness Is Rewarding

                            EXPERIMENT 2                                          would have accepted all offers. This acceptance rate decreased
                                                                                  significantly (prep > .95) as the proportion of the offer relative to
Method                                                                            the stake size decreased (see Table 1). Across all trials in the
Participants and Task                                                             study, multiple participants rejected offers as financially de-
Twelve undergraduates (average age 5 21.8 years; 9 females, 3                     sirable as $8 out of $23; on average, participants rejected at least
males) participated. The task was similar to that in Experiment                   one offer as high as $4.88.
1, except that the stakes ranged from $1 to $23 (average 5
$9.60), and participants underwent fMRI scanning while they                       Self-Report Results
considered the offers. After the scanning session, participants                   Average self-reported estimates of what constituted a fair offer
indicated what they considered a fair offer for each stake size.                  ranged from 45.2% to 48.3% across stakes. For each partici-
                                                                                  pant, we calculated the percentage of unfair offers (as identified
fMRI Acquisition and Analysis                                                     by the participant’s own responses after the scanning session)
Data were acquired on a GE 3-T full-body scanner. Scanning                        that were accepted. On average, participants accepted 49.0% of
parameters were identical to those used in our previous studies                   the offers that were below their self-reported fairness thresholds.
(see Lieberman, Jarcho, & Satpute, 2004). Each of four func-                      These results suggest that although participants were influenced
tional scans consisted of thirty-one 6-s trials, as well as five 6-s              by fairness, they were sometimes able to overcome or disregard
jitter trials. The MR data were analyzed using SPM99 (Wellcome                    fairness considerations and make the economically normative
Department of Cognitive Neurology, London). Images for each                       decision when fairness and material considerations were at
participant were realigned, slice-timed, normalized to Montreal                   odds.
Neurological Institute (MNI) space, and smoothed with an 8-mm
Gaussian kernel, full width at half maximum.                                      fMRI Results: Fairness Preference
   We included in the fairness analysis only matched pairs of                     Several brain regions associated with reward processes were
trials in which the same amount was accepted in one case and                      more active during high-fairness offers than during low-fairness
rejected in the other. Thus, we analyzed neural activation in re-                 offers, after controlling for material utility. Brain regions that
sponse to offers that possessed matched financial rewards and                     showed greater activity for high- than for low-fairness offers
therefore differed primarily in their perceived fairness.                         were the ventral striatum, the amygdala, VMPFC, OFC, and a
   Events were modeled with a canonical hemodynamic re-                           midbrain region near the substantia nigra (see Table 2 and Fig.
sponse function time-locked with the onset of the offer.1 Linear                  3). A post hoc whole-brain analysis also revealed sensitivity to
contrasts were employed to assess comparisons of interest within                  increased fairness in lateral temporal cortex (x 5 44, y 5 18,
individual participants. Random-effects analyses of the group                     z 5 24), t(11) 5 4.74, prep > .99, d 5 2.86.
were computed using the contrast images generated for each                           In the high- versus low-fairness contrast, participants pro-
participant. For regions of interest, significance was set using an               ducing greater activity in the ventral striatum, compared with
uncorrected p value of .005 (5-voxel threshold). Post hoc anal-                   other participants, tended to produce greater activity in the
yses were carried out at a p value of .001 (20-voxel threshold).                  amygdala (r 5 .81, prep > .95) and VMPFC (r 5 .65, prep > .95)
Peristimulus hemodynamic time courses were computed by                            as well. This finding is consistent with the hypothesis that the
identifying clusters of activations from the random-effects                       ventral striatum, the amygdala, and VMPFC function together as
analyses and then applying to these clusters a selective aver-                    a motivational circuit related to reward (Trepel et al., 2005).
aging procedure on a participant-by-participant basis (Pold-                         We examined whether the reward activations could be at-
rack, 2004). Regression analysis of the tendency to accept unfair                 tributed to a continuation of effects occurring during the first
offers was conducted by performing a group analysis in which                      half of each trial, when the face and stake size were presented,
each participant’s rate of accepting unfair offers was entered as a               prior to when the offer was presented and a choice was made. We
regressor to identify which activations correlated with the rate.
                                                                                  TABLE 1
                                                                                  Likelihood of an Offer Being Accepted as a Function of the Ratio
Results                                                                           of the Offer to the Stake Size

Behavioral Results                                                                Ratio of offer to stake                             Acceptance rate
On average, participants accepted 56.3% (SD 5 12.3%) of all                               50%                                              97.9%
the offers in the experiment, a result indicating that they were                        40–49%                                             92.3%
not solely motivated by monetary reward, in which case they                             30–39%                                             75.8%
                                                                                        20–29%                                             44.7%
   1
    These imaging techniques allowed us to determine the blood-oxygenation-             10–19%                                             30.8%
level-dependent signal, an index of neuronal activity, associated with specific         < 10%                                               1.4%
types of offers.

342                                                                                                                                  Volume 19—Number 4
Golnaz Tabibnia, Ajay B. Satpute, and Matthew D. Lieberman

TABLE 2                                                                                    performed a new fairness analysis targeting the 3-s period prior
A Priori Regions Showing Greater Activation in High-Fairness                               to offer onset. No motivational areas active in the fairness analy-
Than in Low-Fairness Trials                                                                sis were active in this new analysis.
                                                       Coordinates        No. of           fMRI Results: Accepting Unfair Offers
Region                            Hemisphere       x        y        z    voxels       t
                                                                                           Our findings for the anterior insula are consistent with those of
Ventral striatum                       Left        6   4         0          7     4.03    Sanfey et al. (2003). Specifically, we observed increased anterior
VMPFC                                  Left       14  32       10         10     4.85
                                                                                           insula activity (relative to a resting baseline) during idiograph-
                                       Left       16  16       16          6     3.83
                                       Right       10  60        4          5     3.82
                                                                                           ically defined unfair trials that were rejected (x 5 36, y 5 18,
Orbitofrontal cortex                   Right       36  36       20         24     4.32    z 5 8), t(11) 5 4.39, prep > .99, d 5 2.95. We also examined
Amygdala                               Left       12 4        24         10     3.81    the neural structures activated when responders overcame fair-
                                       Right       20 12       12         32     4.32    ness concerns and accepted offers they considered unfair. The
Midbrain, SN                           Left       14 20        6          9     4.75    hypothesis that accepted unfair offers activate reward circuitry to a
Note. No activation was greater in low-fairness than in high-fairness trials.
                                                                                           greater extent than rejected unfair offers was not supported by the
The coordinates are from the Montreal Neurological Institute (MNI) atlas.                  data. There was no activity in the ventral striatum, the amygdala, or
Significance was based on an uncorrected p value of .005, with a 5-voxel                   VMPFC (at a liberal statistical threshold of p < .01 uncorrected)
threshold. VMPFC 5 ventromedial prefrontal cortex; SN 5 substantia nigra.
                                                                                           when accepted unfair offers were compared with resting baseline.
                                                                                           However, a large cluster (k 5 834) in right VLPFC was active in
                                                                                           this comparison (x 5 50, y 5 44, z 5 8), t(11) 5 5.39, prep > .99,

                          a            Ventral
                                       Striatum
                                                                                           b
                               VMPFC                            Amygdala                                                            Ventral Striatum
                                                                                                                       0.3                             High Fairness
                                                                                                                       0.2                             Low Fairness
                                                                                           Signal Changes

                                                                                                                       0.1

                                                                                                                       0.0
                                                                                                                                          6s                 12s
                                                                                                              −0.1

                                   Fairness Preference                                                        −0.2
                                                                                                                                    Peristimulus Time

      c                                                                                               d
                                                  Amygdala                                                                                 VMPFC
                         0.3                                             High Fairness                                        0.2                        High Fairness
                         0.2                                             Low Fairness
                                                                                                            Signal Changes

                                                                                                                                                         Low Fairness
       Signal Changes

                                                                                                                              0.1
                         0.1
                          0                                                                                                   0.0
                                                       6s                        12s                                                           6s               12s
                        −0.1
                                                                                                                             −0.1
                        −0.2
                                                                                                                             −0.2
                        −0.3
                                               Peristimulus Time                                                                      Peristimulus Time
      Fig. 3. Ventromedial prefrontal cortex (VMPFC), ventral striatum, and amygdala activation associated with fairness preference. The
      illustration (a) shows the location of clusters with significantly greater activation in response to fair compared with unfair offers. The
      graphs present the time course of activity for fair and unfair trials, relative to a resting baseline, in (b) the ventral striatum, (c) the
      amygdala, and (d) VMPFC. Error bars indicate  1 SE. Along the abscissa, 0 s indicates the onset of the offer (which was 3 s after the trial
      began).

Volume 19—Number 4                                                                                                                                                       343
Fairness Is Rewarding

           a                                                               b
                                                                                                                 1.00
                                                                                                                                     r = −.92
                                                                                                                 0.80

                                                                             Accept Unfair Rate
                                                                                                                 0.60

                                                                                                                 0.40

                                                                                                                 0.20

                          Anterior Insula                                                                        0.00
                                                                                         −1.00          −0.50        0.00         0.50           1.00
                                                                                                  Activity in Left Anterior Insula [-34 20 -6]

           c                                                               d
                                                                                                                 1.00
                                                                                                                                    r = .76
                                                                                                                 0.80

                                                                             Accept Unfair Rate                  0.60

                                                                                                                 0.40

                                                                                                                 0.20

                                                                                                                 0.00
                                                                                        −1.00           −0.50        0.00      0.50              1.00
                                 VLPFC                                                               Activity in Right VLPFC [50 24 8]

         Fig. 4. Brain activation associated with the tendency to accept unfair offers. The illustrations show the location of areas in (a)
         left anterior insula and (c) right ventrolateral prefrontal cortex (right VLPFC) whose activation predicted this tendency. The
         corresponding scatter plots (b and d) depict the correlation between signal change in these areas during accepted relative to
         rejected offers and the rate at which participants accepted offers they later identified as unfair.

d 5 3.25, a finding consistent with the hypothesis that accepting          accepted versus rejected offers. As Figure 4 shows, participants
unfair offers may involve emotion regulation.                              who accepted a higher proportion of idiographically defined
   To further explore the involvement of right VLPFC and the               unfair offers showed a greater increase in activity in right
anterior insula in decisions regarding unfair offers, we specifi-          VLPFC (x 5 50, y 5 24, z 5 8), t(11) 5 3.68, prep > .99, d 5
cally compared trials in which unfair offers were accepted with            2.22, and a greater decrease in activity in the anterior insula on
trials in which unfair offers were rejected. As expected, the left         the left (x 5 34, y 5 20, z 5 6), t(11) 5 7.59, prep > .99,
anterior insula was less active when unfair offers were accepted           d 5 4.58, and right (x 5 32, y 5 22, z 5 10), t(11) 5 6.01,
than when they were rejected (x 5 28, y 5 8, z 5 6), t(11) 5             prep > .99, d 5 3.62, during trials in which offers were accepted
4.06, prep > .99, d 5 2.45. Activity in this insula region was             relative to trials in which offers were rejected. Furthermore,
inversely correlated with right VLPFC activity (x 5 58, y 5 34,            activity in right VLPFC was again inversely correlated with
z 5 10) during trials in which unfair offers were accepted                 activity in the left (r 5 .68, prep > .99) and right (r 5 .86,
(r 5 .77, prep > .99), a finding consistent with the hypothesis           prep > .99) anterior insula. No activity in the ventral striatum, the
that right VLPFC reduces insula activity in such cases.                    amygdala, or VMPFC was positively correlated with the ten-
   We also examined the rate at which participants accepted                dency to accept unfair offers. These findings suggest a prefrontal
offers they rated unfair in the postscanning questionnaire, re-            down-regulation of negative emotional responses during the
gressing this index onto brain activations in the contrast of              process of accepting unfair offers.

344                                                                                                                                        Volume 19—Number 4
Golnaz Tabibnia, Ajay B. Satpute, and Matthew D. Lieberman

   If right VLPFC is involved in the acceptance of unfair offers     more consistent with the former interpretation. Activity in the
by reducing negative affect associated with the anterior insula,     insula mediated the relationship between right VLPFC activity
then the relationship between right VLPFC activity and the rate      and tendency to accept unfair offers, and this finding supports
of accepting unfair offers should be mediated by activity in the     the hypothesis that right VLPFC promotes more normative de-
anterior insula. Indeed, the direct path between right VLPFC         cision making by down-regulating activity in the anterior insula
activity and the rate of accepting unfair offers (b 5 .76, prep >    when unfair offers are considered. Thus, the ability to swallow
.99) was significantly mediated by activity in the left anterior     one’s pride, overcome the insult, and take an unfair offer may
insula (Sobel test: Z 5 2.51, prep > .99). After we controlled for   involve active down-regulation of emotional responses to unfair
activity in this insula region, the remaining path between           treatment.
VLPFC activity and the rate of accepting unfair offers was no           Consistent with the idea that right VLPFC activity drove par-
longer significant (b 5 .25, prep 5 .94).                            ticipants’ decisions to accept unfair offers, studies using repet-
                                                                     itive transcranial magnetic stimulation (rTMS) have identified
                         DISCUSSION                                  the right lateral prefrontal cortex as playing a causal role in ra-
                                                                     tional decision making in the ultimatum game (Knoch, Pascual-
Experiment 1 demonstrated that fairness, but not monetary value,     Leone, Meyer, Treyer, & Fehr, 2006; van’t Wout, Kahn, Sanfey, &
predicted self-reported happiness independently of contempt          Aleman, 2005). Interestingly, in the current study, accepted
ratings. In Experiment 2, nearly all commonly identified reward      unfair offers were not associated with increased activity in re-
areas, including the ventral striatum, the amygdala, VMPFC, and      ward-related regions, which supports the interpretation that
OFC, were associated with fairness preference. Together, these       logical rather than hedonic processes guided these particular
results suggest that individuals react to fairness with a positive   decisions. In the two rTMS studies, transient disruption of
hedonic response, rather than that fairness produces a neutral       function in right dorsolateral prefrontal cortex (right DLPFC)
state and only unfairness produces an affective response. These      interfered with rejection of unfair offers. Although the dorsolat-
results are consistent with previous reports that reward regions     eral region identified in these studies is structurally and func-
such as the striatum, VMPFC, and the amygdala are responsive to      tionally distinct from the VLPFC region identified in the current
cooperative partners and behavior (King-Casas et al., 2005;          study, these findings together are consistent with the notion (van’t
Rilling et al., 2002; Singer, Kiebel, Winston, Dolan, & Frith,       Wout et al., 2005) that the default response in the ultimatum
2004). Further, these results suggest that fairness processing is    game is to reject an unfair offer; DLPFC may be needed
relatively automatic and intuitive, as the ventral striatum, the     to maintain this default goal, whereas VLPFC may be needed to
amygdala, and VMPFC have all been associated with automatic          override it. An alternative interpretation is that DLPFC may
and intuitive processes (Lieberman, 2007).                           be needed to override material self-interest (Knoch et al., 2006),
   In previous studies of the ultimatum game, fairness and ma-       as the role of VLPFC may be to override fairness concerns.
terial utility have covaried. Typical offers in these studies have      In previous studies, right VLPFC has consistently been as-
been either completely fair and of highest material utility (e.g.,   sociated with the down-regulation of activity in regions sup-
$5 out of $10), and therefore easily accepted, or very unfair and    porting negative affect (Lieberman, in press). Specifically, it has
of lowest material utility (e.g., $1 out of $10), and therefore      been associated with the down-regulation of the amygdala’s
usually rejected. However, in real life there are situations in      response to pictures of disturbing scenes (Hariri, Mattay, Tessi-
which the choice is less straightforward. In the current study, we   tore, Fera, & Weinberger, 2003) and pictures of evocative faces
manipulated conflict-ridden choices by presenting offers that        (Hariri et al., 2000; Lieberman et al., 2007; Lieberman, Hariri,
were of considerable material utility but relatively unfair (e.g.,   Jarcho, Eisenberger, & Bookheimer, 2005), as well as the down-
$8 out of $23).                                                      regulation of the anterior cingulate’s response to physical (Lie-
   To the extent that a participant produced increased activity in   berman, Jarcho, Berman, et al., 2004; Wager et al., 2004) and
right VLPFC and decreased activity in the anterior insula, the       social (Eisenberger, Lieberman, & Williams, 2003) pain. Right
participant was more likely to accept unfair but financially de-     VLPFC has also been linked to reduced susceptibility to
sirable offers. One interpretation of this result is that partici-   amygdala-mediated framing effects and thus to rational decision
pants who accepted unfair offers were better able than others to     making (De Martino, Kumaran, Seymour, & Dolan, 2006). Thus,
down-regulate the negative emotional response to unfair treat-       it is plausible that in the current study, right VLPFC down-
ment. Alternatively, increased activity in anterior insula could     regulated affect-related activity in the anterior insula to enable
reflect the ‘‘pain of paying’’ associated with rejecting an offer    the rational decision of accepting an unfair offer.
(Knutson, Rick, Wimmer, Prelec, & Loewenstein, 2007), rather            The fact that we did not find any region exhibiting greater
than the social pain of unfair treatment. Thus, participants who     activity for low-fairness offers than for high-fairness offers
were more likely to accept unfair offers may have had lower          seems at odds with the results of Sanfey et al. (2003), who re-
insula activity because they experienced little pain of paying.      ported greater activity in anterior insula and prefrontal areas
However, the overall pattern of results involving right VLPFC is     during unfair than during fair offers. It is possible that the differ-

Volume 19—Number 4                                                                                                                    345
Fairness Is Rewarding

ences in results are due to differences in the experimental de-        Moreover, when material utility outweighs social utility, people
signs, such as the use of rapid versus slow event-related designs      may down-regulate their affect-related neural response to unfair
or the types of fair and unfair trials used. Nonetheless, in both      treatment in order to choose the economically desirable option.
investigations, increased insula activity correlated with the ten-     These results support the notion that the automatic or default
dency to reject unfair offers, so the studies converge on similar      reaction in economic decision making is to prefer the fair and
functional interpretations of insula activity.                         refuse the unfair (van’t Wout et al., 2005), not just because fair
   One may wonder whether the reward activations observed in           options also tend to be materially advantageous or because
the fairness-preference analysis were in response to fairness per      unfairness is jarring, but also because fair treatment can be
se or were actually due to the perceived higher probability of         rewarding in itself.
receiving the monetary offer in fair trials (which were accepted)
than in unfair trials (which were rejected). Although our data         Acknowledgments—We thank the UCLA Brain Mapping
cannot fully rule out the latter interpretation, it is unlikely. If    Medical Organization, the Ahmanson Foundation, the Pierson-
greater expectation of monetary payoff was driving reward ac-          Lovelace Foundation, and the Tamkin Foundation for support
tivity during high-fairness trials in Experiment 2, we should          and John Monterosso and Daniel Gilbert for helpful comments
have found reward activity during acceptance of unfair offers,         on earlier drafts.
which had the same perceived probability of payoff as the ac-
cepted fair offers. However, activation during accepted unfair                                    REFERENCES
offers was not greater than baseline activation in any reward-
related regions, which suggests that the increased activity in         Baumeister, R.F., & Leary, M.R. (1995). The need to belong: Desire for
these regions during high-fairness trials was not driven by the             interpersonal attachments as a fundamental human motivation.
                                                                            Psychological Bulletin, 117, 497–529.
perceived probability of monetary payoff.
                                                                       Camerer, C., Loewenstein, G., & Prelec, D. (2005). Neuroeconomics:
   Finally, the relatively early onset of activity in the amygdala          How neuroscience can inform economics. Journal of Economic
(see Fig. 3c) suggests that this response may have been both                Literature, 43, 9–64.
related to the offer and a continuation of response from the first     Camerer, C., & Thaler, R.H. (1995). Anomalies: Ultimatums, dictators,
half of the trial. However, we did not find significant amygdala            and manner. Journal of Economic Perspectives, 9, 209–219.
activation in a whole-brain analysis of the first half of the trial,   Cardinal, R.N., Parkinson, J.A., Hall, J., & Everitt, B.J. (2002). Emotion
                                                                            and motivation: The role of the amygdala, ventral striatum, and
even at a lenient threshold (prep > .99, uncorrected). The dis-
                                                                            prefrontal cortex. Neuroscience & Biobehavioral Reviews, 26,
crepancy between this null result and the apparent time course              321–352.
of amygdala activation in Figure 3c reflects the fact that whole-      De Cremer, D., & Alberts, H.J.E.M. (2004). When procedural fairness
brain analyses examine the correlation between the blood-oxy-               does not influence how positive I feel: The effects of voice and
genation-level-dependent response and the canonical hemo-                   leader selection as a function of belongingness need. European
dynamic response, and do not show effects at any one time point.            Journal of Social Psychology, 34, 333–344.
                                                                       De Martino, B., Kumaran, D., Seymour, B., & Dolan, R.J. (2006).
Despite the absence of significant amygdala activation in this
                                                                            Frames, biases, and rational decision-making in the human brain.
analysis, however, it is difficult to infer that the amygdala ac-           Science, 313, 684–687.
tivity reported here reflects reward processing per se, as such        Eisenberger, N.I., Lieberman, M.D., & Williams, K.D. (2003). Does
activity has also been observed in numerous studies of nega-                rejection hurt? An fMRI study of social exclusion. Science, 302,
tively valenced information processing. However, when amyg-                 290–292.
dala activity co-occurs with activity in VMPFC and ventral             Fehr, E., & Schmidt, K.M. (1999). A theory of fairness, competition,
                                                                            and cooperation. Quarterly Journal of Economics, 114, 817–868.
striatum, it is commonly in the context of reward processing           Güth, W., Schmittberger, R., & Schwarze, B. (1982). An experimental
(Cardinal et al., 2002; Petrovich, Holland, & Gallagher, 2005;              analysis of ultimatum bargaining. Journal of Economic Behavior
Trepel et al., 2005). Our confidence in the inference that the              and Organization, 3, 367–388.
amygdala response is related to reward processing is increased         Hariri, A.R., Bookheimer, S.Y., & Mazziotta, J.C. (2000). Modulating
because of (a) the convergence of evidence from self-report and             emotional responses: Effects of a neocortical network on the
                                                                            limbic system. NeuroReport, 11, 43–48.
(b) the fact that this response was observed in the context of
                                                                       Hariri, A.R., Mattay, V.S., Tessitore, A., Fera, F., & Weinberger, D.R.
the activation of multiple regions that work together in a net-             (2003). Neocortical modulation of the amygdala response to
work underlying reward processing (Poldrack, 2006). Future                  fearful stimuli. Biological Psychiatry, 53, 494–501.
studies are needed to elucidate the role of the amygdala in social     Hegtvedt, K.A., & Killian, C. (1999). Fairness and emotions: Reac-
reward.                                                                     tions to the process and outcomes of negotiations. Social Forces,
   In conclusion, these findings suggest that people may prefer             78, 269–303.
                                                                       Hommer, D.W., Knutson, B., Fong, G.W., Bennett, S., Adams, C.M., &
fair outcomes at the cost of material utility in part because they
                                                                            Varnera, J.L. (2003). Amygdalar recruitment during anticipation
hedonically value fairness itself; this preference may not be               of monetary rewards: An event-related fMRI study. In A. Pit-
motivated solely by negative emotional responses to unfairness              kanen, A. Shekhar, & L. Cahill (Eds.), Annals of the New York
or by the impersonal application of culture-driven rules.                   Academy of Sciences: Vol. 985. The amygdala in brain function:

346                                                                                                                          Volume 19—Number 4
Golnaz Tabibnia, Ajay B. Satpute, and Matthew D. Lieberman

     Basic and clinical approaches (pp. 476–478). New York: New           Poldrack, R.A. (2006). Can cognitive processes be inferred from neu-
     York Academy of Sciences.                                                 roimaging data? Trends in Cognitive Sciences, 10, 59–63.
Kahneman, D., Knetsch, J.L., & Thaler, R. (1986). Fairness as a           Rilling, J.K., Gutman, D.A., Zeh, T.R., Pagnoni, G., Berns, G.S., &
     constraint on profit seeking: Entitlements in the market. The             Kilts, C.D. (2002). Neural basis of social cooperation. Neuron, 35,
     American Economic Review, 76, 728–741.                                    395–405.
King-Casas, B., Tomlin, D., Anen, C., Camerer, C.F., Quartz, S.R., &      Sanfey, A.G., Rilling, J.K., Aronson, J.A., Nystrom, L.E., & Cohen,
     Montague, P.R. (2005). Getting to know you: Reputation and trust          J.D. (2003). The neural basis of economic decision-making in the
     in a two-person economic exchange. Science, 308, 78–83.                   Ultimatum Game. Science, 300, 1755–1758.
Knoch, D., Pascual-Leone, A., Meyer, K., Treyer, V., & Fehr, E. (2006).   Sears, D.O., & Funk, C.L. (1991). The role of self-interest in social
     Diminishing reciprocal fairness by disrupting the right prefrontal        and political attitudes. In M.P. Zanna (Ed.), Advances in experi-
     cortex. Science, 314, 829–832.                                            mental social psychology (Vol. 24, pp. 2–91). New York: Aca-
Knutson, B., Rick, S., Wimmer, G.E., Prelec, D., & Loewenstein, G.             demic Press.
     (2007). Neural predictors of purchases. Neuron, 53, 147–156.         Singer, T., Kiebel, S.J., Winston, J.S., Dolan, R.J., & Frith, C.D. (2004).
Lieberman, M.D. (2007). The X- and C-systems: The neural basis of              Brain responses to the acquired moral status of faces. Neuron, 19,
     automatic and controlled social cognition. In E. Harmon-Jones &           653–662.
     P. Winkielman (Eds.), Fundamentals of social neuroscience (pp.       Trepel, C., Fox, C.R., & Poldrack, R.A. (2005). Prospect theory on the
     290–315). New York: Guilford.                                             brain? Toward a cognitive neuroscience of decision under risk.
Lieberman, M.D. (in press). Why symbolic processing of affect can              Brain Research & Cognitive Brain Research, 23, 34–50.
     disrupt negative affect: Social cognitive and affective neu-         Tyler, T.R. (1984). The role of perceived injustice in defendants’
     roscience investigations. In A. Todorov, S.T. Fiske, & D. Prentice        evaluations of their courtroom experience. Law & Society Review,
     (Eds.), Social neuroscience: Toward understanding the underpin-           18, 51–74.
     nings of the social mind. New York: Oxford University Press.         Tyler, T.R. (1991). Psychological models of the justice motive: Ante-
Lieberman, M.D., Eisenberger, N.I., Crockett, M.J., Tom, S.M., Pfeifer,        cedents of distributive and procedural justice. Journal of Per-
     J.H., & Way, B.M. (2007). Putting feelings into words: Affect             sonality and Social Psychology, 67, 850–863.
     labeling disrupts amygdala activity in response to affective stim-   van’t Wout, M., Kahn, R.S., Sanfey, A.G., & Aleman, A. (2005). Re-
     uli. Psychological Science, 18, 421–428.                                  petitive transcranial magnetic stimulation over the right dorso-
Lieberman, M.D., Hariri, A., Jarcho, J.M., Eisenberger, N.I., &                lateral prefrontal cortex affects strategic decision-making.
     Bookheimer, S.Y. (2005). An fMRI investigation of race-related            NeuroReport, 16, 1849–1852.
     amygdala activity in African-American and Caucasian-American         Wager, T.D., Rilling, J.K., Smith, E.E., Sokolik, A., Casey, K.L., Da-
     individuals. Nature Neuroscience, 8, 720–722.                             vidson, R.J., et al. (2004). Placebo-induced changes in fMRI in
Lieberman, M.D., Jarcho, J.M., Berman, S., Naliboff, B., Suyenobu,             the anticipation and experience of pain. Science, 303, 1162–
     B.Y., Mandelkern, M., & Mayer, E.A. (2004). The neural corre-             1167.
     lates of placebo effects: A disruption account. NeuroImage, 22,      Zalla, T., Koechlin, E., Pietrini, P., Basso, G., Aquino, P., Sirigu, A., &
     447–455.                                                                  Grafman, J. (2000). Differential amygdala responses to winning
Lieberman, M.D., Jarcho, J.M., & Satpute, A.B. (2004). Evidence-               and losing: A functional magnetic resonance imaging study in
     based and intuition-based self knowledge: An fMRI study.                  humans. European Journal of Neuroscience, 12, 1764–1770.
     Journal of Personality and Social Psychology, 87, 421–435.
Petrovich, G.D., Holland, P.C., & Gallagher, M. (2005). Amygdalar
     and prefrontal pathways to the lateral hypothalamus are activated
     by a learned cue that stimulates eating. Journal of Neuroscience,
     36, 8295–8302.
Poldrack, R.A. (2004). SPM_ROI_graph. Retrieved August 26, 2004,
     from http://spm-toolbox.sourceforge.net                                         (RECEIVED 3/21/07; REVISION ACCEPTED 9/18/07)

Volume 19—Number 4                                                                                                                              347
You can also read